Compute reference matrix from an `screference` object using scaden (Single-cell assisted deconvolutional network)
scaden_scref.RdCompute reference matrix from an `screference` object using scaden (Single-cell assisted deconvolutional network)
Usage
scaden_scref(
scref,
cache_path = "scaden",
bulk_mat = NULL,
force_retrain = FALSE,
n_cells_sim = 100,
n_samples_sim = 1000,
batch_size = 128,
learning_rate = 1e-04,
steps = 1000,
seed = 0,
gpu = TRUE
)Arguments
- scref
an object of class `screference`
- cache_path
the path to the directory where the intermediate files will be stored (default is "scaden")
- bulk_mat
a matrix containing the bulk gene expression data. If NULL, a pseudobulk will be created from the reference.
- force_retrain
If TRUE, the model will be retrained even if a trained model already exists (default is FALSE)
- n_cells_sim
the number of cells to simulate (default is 100)
- n_samples_sim
the number of samples to simulate (default is 1000)
- batch_size
the batch size used during training (default is 128)
- learning_rate
the learning rate used during training (default is 0.0001)
- steps
the number of training steps (default is 1000)
- seed
the random seed used during training (default is 0)
- gpu
if TRUE, installs Tensorflow and uses GPU acceleration (default is FALSE)